[SNAP-1136] Kryo closure serialtization support and optimizations (#27)

- added back configurable closure serializer in Spark which was removed in SPARK-12414; some minor changes taken from closed Spark PR apache#6361 - added optimized Kryo serialization for multiple classes; currently registration and string sharing fix for kryo (EsotericSoftware/kryo#128) is only in the SnappyData layer PooledKryoSerializer implementation; classes providing maximum benefit have added KryoSerializable notably Accumulators and *Metrics - use closureSerializer for Netty messaging too instead of fixed JavaSerializer - updated kryo to 4.0.0 to get the fix for kryo#342 - actually fixing scalastyle errors introduced by d80ef1b - set ordering field with kryo serialization in GenerateOrdering - removed warning if non-closure passed for cleaning Conflicts: core/src/main/scala/org/apache/spark/network/netty/NettyBlockTransferService.scala core/src/main/scala/org/apache/spark/rpc/netty/NettyRpcEnv.scala core/src/main/scala/org/apache/spark/scheduler/Task.scala core/src/main/scala/org/apache/spark/scheduler/TaskDescription.scala core/src/main/scala/org/apache/spark/scheduler/TaskSetManager.scala core/src/main/scala/org/apache/spark/storage/BlockId.scala core/src/main/scala/org/apache/spark/util/AccumulatorV2.scala sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/codegen/GenerateSafeProjection.scala sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/codegen/GenerateUnsafeProjection.scala sql/core/src/main/scala/org/apache/spark/sql/execution/metric/SQLMetrics.scala
TIBCOSoftware · Mar 5, 2018 · b500538 · b500538
1 parent 59d8076
commit b500538
Show file tree

Hide file tree

Showing 29 changed files with 666 additions and 121 deletions.
diff --git a/build.gradle b/build.gradle
@@ -61,7 +61,8 @@ allprojects {
     javaxServletVersion = '3.1.0'
     guavaVersion = '14.0.1'
     hiveVersion = '1.2.1.spark2'
-    chillVersion = '0.8.0'
+    chillVersion = '0.8.1'
+    kryoVersion = '4.0.0'
     nettyVersion = '3.8.0.Final'
     nettyAllVersion = '4.0.29.Final'
     derbyVersion = '10.12.1.1'

diff --git a/common/unsafe/build.gradle b/common/unsafe/build.gradle
@@ -20,7 +20,10 @@ description = 'Spark Project Unsafe'
 dependencies {
   compile project(subprojectBase + 'snappy-spark-tags_' + scalaBinaryVersion)
 
-  compile group: 'com.twitter', name: 'chill_' + scalaBinaryVersion, version: chillVersion
+  compile group: 'com.esotericsoftware', name: 'kryo-shaded', version: kryoVersion
+  compile(group: 'com.twitter', name: 'chill_' + scalaBinaryVersion, version: chillVersion) {
+    exclude(group: 'com.esotericsoftware', module: 'kryo-shaded')
+  }
   compile group: 'com.google.code.findbugs', name: 'jsr305', version: jsr305Version
   compile group: 'com.google.guava', name: 'guava', version: guavaVersion
 

diff --git a/core/build.gradle b/core/build.gradle
@@ -40,8 +40,13 @@ dependencies {
     exclude(group: 'org.apache.avro', module: 'avro-ipc')
   }
   compile group: 'com.google.guava', name: 'guava', version: guavaVersion
-  compile group: 'com.twitter', name: 'chill_' + scalaBinaryVersion, version: chillVersion
-  compile group: 'com.twitter', name: 'chill-java', version: chillVersion
+  compile group: 'com.esotericsoftware', name: 'kryo-shaded', version: kryoVersion
+  compile(group: 'com.twitter', name: 'chill_' + scalaBinaryVersion, version: chillVersion) {
+    exclude(group: 'com.esotericsoftware', module: 'kryo-shaded')
+  }
+  compile(group: 'com.twitter', name: 'chill-java', version: chillVersion) {
+    exclude(group: 'com.esotericsoftware', module: 'kryo-shaded')
+  }
   compile group: 'org.apache.xbean', name: 'xbean-asm5-shaded', version: '4.4'
   // explicitly include netty from akka-remote to not let zookeeper override it
   compile group: 'io.netty', name: 'netty', version: nettyVersion

diff --git a/core/src/main/scala/org/apache/spark/SparkEnv.scala b/core/src/main/scala/org/apache/spark/SparkEnv.scala
@@ -170,6 +170,43 @@ object SparkEnv extends Logging {
     env
   }
 
+  // Create an instance of the class with the given name, possibly initializing it with our conf
+  def instantiateClass[T](className: String, conf: SparkConf,
+      isDriver: Boolean): T = {
+    val cls = Utils.classForName(className)
+    // Look for a constructor taking a SparkConf and a boolean isDriver, then one taking just
+    // SparkConf, then one taking no arguments
+    try {
+      cls.getConstructor(classOf[SparkConf], java.lang.Boolean.TYPE)
+          .newInstance(conf, new java.lang.Boolean(isDriver))
+          .asInstanceOf[T]
+    } catch {
+      case _: NoSuchMethodException =>
+        try {
+          cls.getConstructor(classOf[SparkConf]).newInstance(conf).asInstanceOf[T]
+        } catch {
+          case _: NoSuchMethodException =>
+            cls.getConstructor().newInstance().asInstanceOf[T]
+        }
+    }
+  }
+
+  def getClosureSerializer(conf: SparkConf, doLog: Boolean = false): Serializer = {
+    val defaultClosureSerializerClass = classOf[JavaSerializer].getName
+    val closureSerializerClass = conf.get("spark.closure.serializer",
+      defaultClosureSerializerClass)
+    val closureSerializer = instantiateClass[Serializer](
+      closureSerializerClass, conf, isDriver = false)
+    if (doLog) {
+      if (closureSerializerClass != defaultClosureSerializerClass) {
+        logInfo(s"Using non-default closure serializer: $closureSerializerClass")
+      } else {
+        logDebug(s"Using closure serializer: $closureSerializerClass")
+      }
+    }
+    closureSerializer
+  }
+
   /**
    * Create a SparkEnv for the driver.
    */
@@ -272,26 +309,9 @@ object SparkEnv extends Logging {
       conf.set("spark.driver.port", rpcEnv.address.port.toString)
     }
 
-    // Create an instance of the class with the given name, possibly initializing it with our conf
     def instantiateClass[T](className: String): T = {
-      val cls = Utils.classForName(className)
-      // Look for a constructor taking a SparkConf and a boolean isDriver, then one taking just
-      // SparkConf, then one taking no arguments
-      try {
-        cls.getConstructor(classOf[SparkConf], java.lang.Boolean.TYPE)
-          .newInstance(conf, new java.lang.Boolean(isDriver))
-          .asInstanceOf[T]
-      } catch {
-        case _: NoSuchMethodException =>
-          try {
-            cls.getConstructor(classOf[SparkConf]).newInstance(conf).asInstanceOf[T]
-          } catch {
-            case _: NoSuchMethodException =>
-              cls.getConstructor().newInstance().asInstanceOf[T]
-          }
-      }
+      SparkEnv.instantiateClass(className, conf, isDriver)
     }
-
     // Create an instance of the class named by the given SparkConf property, or defaultClassName
     // if the property is not set, possibly initializing it with our conf
     def instantiateClassFromConf[T](propertyName: String, defaultClassName: String): T = {
@@ -304,7 +324,7 @@ object SparkEnv extends Logging {
 
     val serializerManager = new SerializerManager(serializer, conf, ioEncryptionKey)
 
-    val closureSerializer = new JavaSerializer(conf)
+    val closureSerializer = getClosureSerializer(conf, doLog = true)
 
     def registerOrLookupEndpoint(
         name: String, endpointCreator: => RpcEndpoint):

diff --git a/core/src/main/scala/org/apache/spark/executor/InputMetrics.scala b/core/src/main/scala/org/apache/spark/executor/InputMetrics.scala
@@ -17,6 +17,10 @@
 
 package org.apache.spark.executor
 
+import com.esotericsoftware.kryo.{Kryo, KryoSerializable}
+import com.esotericsoftware.kryo.io.{Input, Output}
+
+import org.apache.spark.TaskContext
 import org.apache.spark.annotation.DeveloperApi
 import org.apache.spark.util.LongAccumulator
 
@@ -39,7 +43,7 @@ object DataReadMethod extends Enumeration with Serializable {
  * A collection of accumulators that represents metrics about reading data from external systems.
  */
 @DeveloperApi
-class InputMetrics private[spark] () extends Serializable {
+class InputMetrics private[spark] () extends Serializable with KryoSerializable {
   private[executor] val _bytesRead = new LongAccumulator
   private[executor] val _recordsRead = new LongAccumulator
 
@@ -56,4 +60,18 @@ class InputMetrics private[spark] () extends Serializable {
   private[spark] def incBytesRead(v: Long): Unit = _bytesRead.add(v)
   private[spark] def incRecordsRead(v: Long): Unit = _recordsRead.add(v)
   private[spark] def setBytesRead(v: Long): Unit = _bytesRead.setValue(v)
+
+  override def write(kryo: Kryo, output: Output): Unit = {
+    _bytesRead.write(kryo, output)
+    _recordsRead.write(kryo, output)
+  }
+
+  override final def read(kryo: Kryo, input: Input): Unit = {
+    read(kryo, input, context = null)
+  }
+
+  def read(kryo: Kryo, input: Input, context: TaskContext): Unit = {
+    _bytesRead.read(kryo, input, context)
+    _recordsRead.read(kryo, input, context)
+  }
 }
diff --git a/core/src/main/scala/org/apache/spark/executor/OutputMetrics.scala b/core/src/main/scala/org/apache/spark/executor/OutputMetrics.scala
@@ -17,6 +17,10 @@
 
 package org.apache.spark.executor
 
+import com.esotericsoftware.kryo.{Kryo, KryoSerializable}
+import com.esotericsoftware.kryo.io.{Input, Output}
+
+import org.apache.spark.TaskContext
 import org.apache.spark.annotation.DeveloperApi
 import org.apache.spark.util.LongAccumulator
 
@@ -38,7 +42,7 @@ object DataWriteMethod extends Enumeration with Serializable {
  * A collection of accumulators that represents metrics about writing data to external systems.
  */
 @DeveloperApi
-class OutputMetrics private[spark] () extends Serializable {
+class OutputMetrics private[spark] () extends Serializable with KryoSerializable {
   private[executor] val _bytesWritten = new LongAccumulator
   private[executor] val _recordsWritten = new LongAccumulator
 
@@ -54,4 +58,18 @@ class OutputMetrics private[spark] () extends Serializable {
 
   private[spark] def setBytesWritten(v: Long): Unit = _bytesWritten.setValue(v)
   private[spark] def setRecordsWritten(v: Long): Unit = _recordsWritten.setValue(v)
+
+  override def write(kryo: Kryo, output: Output): Unit = {
+    _bytesWritten.write(kryo, output)
+    _recordsWritten.write(kryo, output)
+  }
+
+  override final def read(kryo: Kryo, input: Input): Unit = {
+    read(kryo, input, context = null)
+  }
+
+  def read(kryo: Kryo, input: Input, context: TaskContext): Unit = {
+    _bytesWritten.read(kryo, input, context)
+    _recordsWritten.read(kryo, input, context)
+  }
 }
diff --git a/core/src/main/scala/org/apache/spark/executor/ShuffleReadMetrics.scala b/core/src/main/scala/org/apache/spark/executor/ShuffleReadMetrics.scala
@@ -17,6 +17,10 @@
 
 package org.apache.spark.executor
 
+import com.esotericsoftware.kryo.{Kryo, KryoSerializable}
+import com.esotericsoftware.kryo.io.{Input, Output}
+
+import org.apache.spark.TaskContext
 import org.apache.spark.annotation.DeveloperApi
 import org.apache.spark.util.{DoubleAccumulator, LongAccumulator}
 
@@ -27,7 +31,7 @@ import org.apache.spark.util.{DoubleAccumulator, LongAccumulator}
  * Operations are not thread-safe.
  */
 @DeveloperApi
-class ShuffleReadMetrics private[spark] () extends Serializable {
+class ShuffleReadMetrics private[spark] () extends Serializable with KryoSerializable {
   private[executor] val _remoteBlocksFetched = new LongAccumulator
   private[executor] val _localBlocksFetched = new LongAccumulator
   private[executor] val _remoteBytesRead = new LongAccumulator
@@ -121,6 +125,28 @@ class ShuffleReadMetrics private[spark] () extends Serializable {
       _recordsRead.add(metric.recordsRead)
     }
   }
+
+  override def write(kryo: Kryo, output: Output): Unit = {
+    _remoteBlocksFetched.write(kryo, output)
+    _localBlocksFetched.write(kryo, output)
+    _remoteBytesRead.write(kryo, output)
+    _localBytesRead.write(kryo, output)
+    _fetchWaitTime.write(kryo, output)
+    _recordsRead.write(kryo, output)
+  }
+
+  override final def read(kryo: Kryo, input: Input): Unit = {
+    read(kryo, input, context = null)
+  }
+
+  def read(kryo: Kryo, input: Input, context: TaskContext): Unit = {
+    _remoteBlocksFetched.read(kryo, input, context)
+    _localBlocksFetched.read(kryo, input, context)
+    _remoteBytesRead.read(kryo, input, context)
+    _localBytesRead.read(kryo, input, context)
+    _fetchWaitTime.read(kryo, input, context)
+    _recordsRead.read(kryo, input, context)
+  }
 }
 
 /**

diff --git a/core/src/main/scala/org/apache/spark/executor/ShuffleWriteMetrics.scala b/core/src/main/scala/org/apache/spark/executor/ShuffleWriteMetrics.scala
@@ -17,6 +17,10 @@
 
 package org.apache.spark.executor
 
+import com.esotericsoftware.kryo.{Kryo, KryoSerializable}
+import com.esotericsoftware.kryo.io.{Input, Output}
+
+import org.apache.spark.TaskContext
 import org.apache.spark.annotation.DeveloperApi
 import org.apache.spark.util.LongAccumulator
 
@@ -27,7 +31,7 @@ import org.apache.spark.util.LongAccumulator
  * Operations are not thread-safe.
  */
 @DeveloperApi
-class ShuffleWriteMetrics private[spark] () extends Serializable {
+class ShuffleWriteMetrics private[spark] () extends Serializable with KryoSerializable {
   private[executor] val _bytesWritten = new LongAccumulator
   private[executor] val _recordsWritten = new LongAccumulator
   private[executor] val _writeTime = new LongAccumulator
@@ -57,6 +61,22 @@ class ShuffleWriteMetrics private[spark] () extends Serializable {
     _recordsWritten.setValue(recordsWritten - v)
   }
 
+  override def write(kryo: Kryo, output: Output): Unit = {
+    _bytesWritten.write(kryo, output)
+    _recordsWritten.write(kryo, output)
+    _writeTime.write(kryo, output)
+  }
+
+  override def read(kryo: Kryo, input: Input): Unit = {
+    read(kryo, input, context = null)
+  }
+
+  def read(kryo: Kryo, input: Input, context: TaskContext): Unit = {
+    _bytesWritten.read(kryo, input, context)
+    _recordsWritten.read(kryo, input, context)
+    _writeTime.read(kryo, input, context)
+  }
+
   // Legacy methods for backward compatibility.
   // TODO: remove these once we make this class private.
   @deprecated("use bytesWritten instead", "2.0.0")

diff --git a/core/src/main/scala/org/apache/spark/executor/TaskMetrics.scala b/core/src/main/scala/org/apache/spark/executor/TaskMetrics.scala
@@ -20,6 +20,9 @@ package org.apache.spark.executor
 import scala.collection.JavaConverters._
 import scala.collection.mutable.{ArrayBuffer, LinkedHashMap}
 
+import com.esotericsoftware.kryo.{Kryo, KryoSerializable}
+import com.esotericsoftware.kryo.io.{Input, Output}
+
 import org.apache.spark._
 import org.apache.spark.annotation.DeveloperApi
 import org.apache.spark.internal.Logging
@@ -42,7 +45,7 @@ import org.apache.spark.util._
  * be sent to the driver.
  */
 @DeveloperApi
-class TaskMetrics private[spark] () extends Serializable {
+class TaskMetrics private[spark] () extends Serializable with KryoSerializable {
   // Each metric is internally represented as an accumulator
   private val _executorDeserializeTime = new DoubleAccumulator
   private val _executorDeserializeCpuTime = new LongAccumulator
@@ -263,6 +266,40 @@ class TaskMetrics private[spark] () extends Serializable {
     // value will be updated at driver side.
     internalAccums.filter(a => !a.isZero || a == _resultSize)
   }
+
+  override def write(kryo: Kryo, output: Output): Unit = {
+    _executorDeserializeTime.write(kryo, output)
+    _executorRunTime.write(kryo, output)
+    _resultSize.write(kryo, output)
+    _jvmGCTime.write(kryo, output)
+    _resultSerializationTime.write(kryo, output)
+    _memoryBytesSpilled.write(kryo, output)
+    _diskBytesSpilled.write(kryo, output)
+    _peakExecutionMemory.write(kryo, output)
+    _updatedBlockStatuses.write(kryo, output)
+    inputMetrics.write(kryo, output)
+    outputMetrics.write(kryo, output)
+    shuffleReadMetrics.write(kryo, output)
+    shuffleWriteMetrics.write(kryo, output)
+  }
+
+  override def read(kryo: Kryo, input: Input): Unit = {
+    // read the TaskContext thread-local once
+    val taskContext = TaskContext.get()
+    _executorDeserializeTime.read(kryo, input, taskContext)
+    _executorRunTime.read(kryo, input, taskContext)
+    _resultSize.read(kryo, input, taskContext)
+    _jvmGCTime.read(kryo, input, taskContext)
+    _resultSerializationTime.read(kryo, input, taskContext)
+    _memoryBytesSpilled.read(kryo, input, taskContext)
+    _diskBytesSpilled.read(kryo, input, taskContext)
+    _peakExecutionMemory.read(kryo, input, taskContext)
+    _updatedBlockStatuses.read(kryo, input, taskContext)
+    inputMetrics.read(kryo, input, taskContext)
+    outputMetrics.read(kryo, input, taskContext)
+    shuffleReadMetrics.read(kryo, input, taskContext)
+    shuffleWriteMetrics.read(kryo, input, taskContext)
+  }
 }
 
 

diff --git a/core/src/main/scala/org/apache/spark/network/netty/NettyBlockTransferService.scala b/core/src/main/scala/org/apache/spark/network/netty/NettyBlockTransferService.scala
@@ -24,9 +24,7 @@ import scala.collection.JavaConverters._
 import scala.concurrent.{Future, Promise}
 import scala.reflect.ClassTag
 
-import com.codahale.metrics.{Metric, MetricSet}
-
-import org.apache.spark.{SecurityManager, SparkConf}
+import org.apache.spark.{SecurityManager, SparkConf, SparkEnv}
 import org.apache.spark.network._
 import org.apache.spark.network.buffer.ManagedBuffer
 import org.apache.spark.network.client.{RpcResponseCallback, TransportClientBootstrap, TransportClientFactory}
@@ -35,7 +33,6 @@ import org.apache.spark.network.server._
 import org.apache.spark.network.shuffle.{BlockFetchingListener, OneForOneBlockFetcher, RetryingBlockFetcher, TempFileManager}
 import org.apache.spark.network.shuffle.protocol.UploadBlock
 import org.apache.spark.network.util.JavaUtils
-import org.apache.spark.serializer.JavaSerializer
 import org.apache.spark.storage.{BlockId, StorageLevel}
 import org.apache.spark.util.Utils
 
@@ -52,7 +49,7 @@ private[spark] class NettyBlockTransferService(
   extends BlockTransferService {
 
   // TODO: Don't use Java serialization, use a more cross-version compatible serialization format.
-  private val serializer = new JavaSerializer(conf)
+  private val serializer = SparkEnv.getClosureSerializer(conf)
   private val authEnabled = securityManager.isAuthenticationEnabled()
   private val transportConf = SparkTransportConf.fromSparkConf(conf, "shuffle", numCores)
 

diff --git a/core/src/main/scala/org/apache/spark/rdd/RDD.scala b/core/src/main/scala/org/apache/spark/rdd/RDD.scala
@@ -147,7 +147,8 @@ abstract class RDD[T: ClassTag](
   def sparkContext: SparkContext = sc
 
   /** A unique ID for this RDD (within its SparkContext). */
-  val id: Int = sc.newRddId()
+  protected var _id: Int = sc.newRddId()
+  def id: Int = _id
 
   /** A friendly name for this RDD */
   @transient var name: String = _
@@ -1651,7 +1652,7 @@ abstract class RDD[T: ClassTag](
   // Other internal methods and fields
   // =======================================================================
 
-  private var storageLevel: StorageLevel = StorageLevel.NONE
+  protected var storageLevel: StorageLevel = StorageLevel.NONE
 
   /** User code that created this RDD (e.g. `textFile`, `parallelize`). */
   @transient private[spark] val creationSite = sc.getCallSite()